Collaborative researches of three-dimensional molecular modeling can be limited by different time zones and locations. A networked virtual environment can be utilized to overcome the problem caused by the temporal and spatial differences. However, traditional approaches did not sufficiently consider integration of different computing environments, which were characterized by types of applications, roles of users, and so on. We propose a collaborative molecular modeling environment to integrate different molecule modeling systems using a virtual tunneling service. We integrated Co-Coot, which is a collaborative crystallographic object-oriented toolkit, with VRMMS, which is a virtual reality molecular modeling system, through a collaborative tunneling system. The proposed system showed reliable quantitative and qualitative results through pilot experiments.
A networked virtual environment (NVE) is a computer-supported collaborative work (CSCW) environment where multiple participants can interact with each other through computer networks for enhancing performances of their collaborations. Since, NVE can overcome limit of time and space differences during face-to-face collaborations, NVE has been widely researched for collaborative computer-aided design [
In order to design new materials and new drugs, we need to understand functions of proteins through analysis of a 3D protein structure at atomic resolution. First, it is generally determined by X-ray crystallography or NMR. Second, we can simulate the behaviors of the 3D molecules with equations of quantum and physics through computer simulations. Third, the 3D model of an enzyme, which is a candidate material or drug, can be used to design a higher binding affinity inhibitor against a target enzyme. Last, we can simulate the designed enzyme to have better characteristics such as higher activity and stability for industrial purpose [
Since molecular modeling is a large and complicated process, its participants generally collaborate together according to their roles and familiar applications. If all participants use a same molecular system, it would work well. However, it is not a case in a real world. The participating biologists generally use their own favorite molecular modeling systems among many available systems. Therefore, real-time collaborations of the participating biologists using NVE systems are not smoothly realized in general.
In this paper, we propose a collaborative experiment environment with different molecular modeling systems. The environment consists of two collaborative systems, VRMMS (virtual reality molecular modeling system) [
There have been several researches on collaborative molecular modeling systems for studying and analyzing 3D biomolecular structures. BioCore [
In our previous papers, VRMMS was suggested to visualize 3D biomolecular structures and calculate simulation of energy minimization [
Coot is open-source software and popularly used. It is specialized to edit 3D biomolecular structures for finding a best enzyme model in a crystallography process [
We designed a system architecture of the collaborative environment as a client/server network topology. It is easy to manage collaborative communications among clients and a server. The proposed environment can be illustrated as shown in Figure
System overview.
The participating researchers can perform their collaboration simultaneously to share their intermediate experimental results, discuss a future direction, or solve their difficult problems via the proposed networked virtual environment. The proposed environment offers a pessimistic concurrency control mechanism, which allows accesses and manipulates shared 3D molecular models with permission from the collaborative server. During the collaborative works, any user can request the collaborative server for his/her authority over a shared molecular model. The server may allow the current request for an authority if it is available. If other user already owns the shared molecular model, the server denies the request from the user. This mechanism could avoid conflicts among multiple requests from the participating researchers.
The proposed networked virtual environment also provides a private workspace if the denied user still wants to manipulate the shared molecular model privately. With the private workspace, the user can manipulate every features of the biomolecular model and the intermediate result can be stored in a file after using the private workspace.
See Figure
Collaborative tunneling service.
In order to provide transformations between two different applications, we need to define a mapping mechanism between the applications. The proposed system provides transformation strategies for the mapping mechanism. We analyzed both VRMMS and Co-Coot in order to extract important functions in a collaborative 3D molecular modeling process. We modeled six functions of VRMMS and Co-Coot as described in Tables
Supporintg functions in VRMMS.
Functions | Detailed operations |
---|---|
Manipulation | Translation, rotation, and scaling |
Rendering | Surface, wire frame, and ball and stick |
Simulation | Energy minimization |
Ownership | Pessimistic concurrency control |
File transferring | File sharing |
Chatting | Exchange text messages |
Supporintg functions in Co-Coot.
Functions | Detailed operations |
---|---|
Manipulation | Translation, rotation, and scaling |
Rendering | Wire frame |
Editing | Add, delete, and rotate an amino acid with bonds |
Ownership | Pessimistic concurrency control |
File transferring | File sharing |
Chatting | Exchange text messages |
Function mapping between VRMMS and Co-Coot.
VRMMS functions | Co-Coot functions | Semantic mapping |
---|---|---|
Manipulation | Manipulation | Equal |
Rendering | Rendering | Convertible |
Simulation | — | Readable |
— | Editing | Readable |
Ownership | Ownership | Equal |
File transferring | File Transferring | Equal |
Chatting | Chatting | Equal |
A user of VRMMS can monitor 3D molecular models with various visualization methods such as wire frame, ball and stick, and surface modes in rendering function. VRMMS also provides a simulation function to calculate energy values of the 3D molecular models. In Co-Coot, a user can visualize molecular models as a wire frame mode. Co-Coot provides an editing function to refine current 3D molecular models. Figure
Results of visualization of 1SFO [
The “rendering” functions of the two applications are semantically and operationally different. But they are “convertible”. So, a “surface” rendering model in VRMMS can be converted and expressed as a “wireframe” model in Co-Coot. The same conversion strategy can be applied to the “ball and stick” rendering model of VRMMS.
Though VRMMS has the “simulation” function, Co-Coot does not, on the other hand. Then, the simulation results from VRMMS could be translated into a text format and read by the Co-Coot users as shown in Figure
Results of editing operations: (a) an original model on Co-Coot, (b) a changed model on Co-Coot, (c) the original model on VRMMS, and (d) the changed model on VRMMS.
We evaluated performance of the proposed environment by several quantitative and qualitative measures. As described in Table
Selected molecular models.
PDB CODE | Name | Number of atoms |
---|---|---|
1SFO | Yeast polymerase II | 28,649 |
First, we compared the rendering speed of Co-Coot and VRMMS with various visualizations using the same molecular models. The experiment was conducted on a desktop PC with Core 2 Quad CPU and an nVdia GTX 265 graphic card. We tested rendering speeds of Co-Coot and VRMMS for 100 seconds in terms of the FPS (frames per second) values with four possible visualization modes. As shown in Figure
Results of various visualization tests: (A) wire frame mode in VRMMS (40.9 fps), (B) ball and stick mode in VRMMS (22.8 fps), (C) wire frame mode in Co-Coot (18.8 fps), and (D) surface mode in VRMMS (16.4 fps).
Second, we measured the number of transformations and average delivering times. In order to measure the values, we conducted a collaborative molecular design process within different environments through network connections as shown in Figure
Collaborative molecular modeling: (a) a user is using Co-Coot; (b) another user is using VRMMS.
Figure
Results of trcollaborative tunneling test: (a) numbers of transformed data according to transformer strategies; (b) average delivering times.
We conducted user interviews with the two subjects after conducting the previous experiments. The participants answered that the proposed environment was generally satisfied. However, they required more collaborative tunneling services such as a voice chatting and a sharing movie. They also answered if the proposed environment could be applied to support other famous molecular modeling systems, then it would be feasible to utilize the proposed environment in the real collaborative molecular modeling processes.
In this paper, we propose a new collaborative molecular modeling environment to connect different modeling systems based on an approach with a collaborative tunneling service and transformation strategies. With our approach, multiple users can collaborative together even though they are manipulating different modeling systems.
The proposed environment showed feasible rendering performances in target applications with various visualizations. The results of another performance evaluation of the pilot tests showed that the proposed environment could successfully transform collaboration data with stable delivering times through network.
An additional user study showed that the participants would like to adopt the proposed system in their collaborative molecular modeling environment. For our future works, we will expand the tunneling services to the other popular molecular modeling systems using ontology to support semantic transformation strategies. We also plan to enhance a concurrency control mechanisms with different roles and applications.
This paper was supported by Konkuk University in 2011.